Life In 19x19
http://lifein19x19.com/

Practical alternatives to superko
http://lifein19x19.com/viewtopic.php?f=45&t=13382		 Page 2 of 2
Author:  DrStraw [ Tue Jul 19, 2016 4:12 am ]
Post subject:  Re: Practical alternatives to superko
Magicwand wrote:
can people apply that idea to triple+ ko or other situation without getting confused?
i think even professionals will have hard time remembering all prior position.

Chochihun even for that that it was his turn to take ko during championship match.
now what would happen if it is triple or quadruple ko. only alpha-go will be able to trace such sequence.



That is exactly why I don't like the superko rule. Imagine telling a complete beginner that he cannot play a point because it reproduces a previous position. He may understand the concept but will not be able to recall the previous position.
Author:  hyperpape [ Tue Jul 19, 2016 7:33 am ]
Post subject:  Re: Practical alternatives to superko
While "greatly" is subjective, I would argue that the 2-2-2 and 2-2 variants are not small changes--they prohibit what go players would perceive to be utterly ordinary ko threats, as in my example (it could be easily modified for the 2-2 rule). Without trying to create a technical definition, I would say that intuitively, a substitute ko rule should never prohibit a placement in an "unrelated" part of the board--unless perhaps that unrelated part of the board is part of a second ko fight. Any rule that fails that condition stops being a refinement of the ko rule and becomes a new go variant (this latter distinction is also fuzzy).

Now, go variants aren't the worst thing in the world. Redstone, for instance, seemed pretty interesting. There might even be some go variant in the space of possible games that would have been a better game than go. But given the fact that I've spent years learning the actual game of go, and the complications around the ko rule are manageable, I'm actually far less interested in go variants than I am in new games that do not have any particularly close relation to go.
Author:  luigi [ Tue Jul 19, 2016 8:17 am ]
Post subject:  Re: Practical alternatives to superko
hyperpape wrote:
While "greatly" is subjective, I would argue that the 2-2-2 and 2-2 variants are not small changes--they prohibit what go players would perceive to be utterly ordinary ko threats, as in my example (it could be easily modified for the 2-2 rule).

The 2-2 rule is more restrictive than the 2-2-2 rule, so your example is valid for both. Of course, one can always refine those rules by having them apply only to sequences of two disturbing captures and by stating that the move after such sequences can be anything except a disturbing play (to be defined as a move with all common features of all capturing and non-capturing moves in all forced cycles). That brings the game further closer to Go and complies better with your requirement for substitute ko rules.

Quote:
Without trying to create a technical definition, I would say that intuitively, a substitute ko rule should never prohibit a placement in an "unrelated" part of the board--unless perhaps that unrelated part of the board is part of a second ko fight.

This is what the "disturbing capture" variant aims to accomplish. Not all disturbing captures will be part of ko fights, but all captures that are part of ko fights will be disturbing captures, since they are defined by the shared features of all captures that are part of ko fights.

Quote:
Any rule that fails that condition stops being a refinement of the ko rule and becomes a new go variant (this latter distinction is also fuzzy)

I think I've always presented Stoical Go as a Go variant, and I agree that these new ones are too. But I also think they are refinements of the ko and superko rules. :)
Author:  Pio2001 [ Tue Jul 19, 2016 12:22 pm ]
Post subject:  Re: Practical alternatives to superko
DrStraw wrote:
There is a practical alternative to superko and it was in use for hundreds of years until someone came up with the, in my opinion, stupid idea of superko. That idea is No Result.


Hi,
I disagree here. The "no result" is even worse than "impractical", it is just impossible to apply in real life.

Imagine the final game of your national go championship. The winner of the game will be this years' go champion of your country.
You have rent a hall for the championship and must return the key in the evening. As usual, the planning of the championship is late.
The two players have their plane booked for the next morning to go home. For the night, they sleep at local player's homes.

After two hours of play, the game ends with no result. What do you do ?

a) Tell the basketball team using the hall after you that they must cancel their championship because you need the hall to replay the game.
b) Have the two players replay the final in your living room and tell them not to pay attention to your nephew who is yelling at the TV set watching the basketball championship.
c) Tell your federation that there will be no champion this year.
d) Having the match replayed another day, but too late to qualify your champion for the World Amateur Championship.
e) Rely on the high probability that no national final ever will end with "no result".


My opinion on the topic is that the the opposition of the simplicity of the rules vs the complexity of the game is an essential part of the beauty of go. Therefore my preferred option is just "no repetition allowed" (i.e. positional superko). This is the simplest rule.

My choice is also based on the fact that a rule should be short and easy to understand. I have always found very odd that no go books for beginner have a chapter called "rules of play".

Remembering repetitions is much easier than playing a long life ko, or even a bent four. For a triple ko, you play 5 captures, a ko threat exchange, 5 captures, a ko threat exchange etc. From a tactical point of view, there is no difference between a triple ko and a simple ko. The player who has more ko threats wins the triple ko.
I don't know how a quadruple ko behaves, though.

According to the statistics in the Go Player's Almanach, between 1960 and 1995, 12 games ended with no result, that is one out of 7300.
Out of these 12 games, 5 were triple ko, 5 were quadruple ko, 1 was eternal life, and 1 was an adjudication.
Author:  oren [ Tue Jul 19, 2016 12:34 pm ]
Post subject:  Re: Practical alternatives to superko
Pio2001 wrote:
DrStraw wrote:
There is a practical alternative to superko and it was in use for hundreds of years until someone came up with the, in my opinion, stupid idea of superko. That idea is No Result.


Hi,
I disagree here. The "no result" is even worse than "impractical", it is just impossible to apply in real life.


No result has been applied in real life many times. My favorite was a tv championship that ended that way. The players got a short break, change of officials and tv commentator, and they restarted the game.

That was easy...

For longer games, you can decide if you want ties or if you want a rescheduled match. Saying it's impractical is a bit silly.
Author:  luigi [ Tue Jul 19, 2016 12:46 pm ]
Post subject:  Re: Practical alternatives to superko
I realize that my variants aren't quite Go, so it was to be expected that Go players wouldn't quite approve of them.

At this moment, my favorite way to handle repetitions in a way that the resulting game is still Go (i.e. in a way that changes only the way long cycles behave and nothing else, just as superko rules do) is Jasiek's Basic-Fixed-Ko rule. The effort needed to recognize repetitions is greater here than it is under the no-result rule (since you can only play out cycles once as opposed to infinitely many times) but lower than it is under superko (where no cycle can ever be played out to the end). And, most importantly, groups involved in would-be cycles are just alive, which seems adequate as you can never capture them under traditional rules (with no superko). So instead of voiding the game, you can just play on without worrying about those groups anymore. This further reduces the amount of mental effort spent in handling repetitions compared to superko.

It seems the perfect compromise between no result and superko to me. Out of curiosity, for those of you who dislike superko: what is your stance on that rule?
Author:  Magicwand [ Thu Jul 21, 2016 8:30 pm ]
Post subject:  Re: Practical alternatives to superko
luigi wrote:
So instead of voiding the game, you can just play on without worrying about those groups anymore. This further reduces the amount of mental effort spent in handling repetitions compared to superko.

It seems the perfect compromise between no result and superko to me. Out of curiosity, for those of you who dislike superko: what is your stance on that rule?

Sorry about being blunt but
Ary you Triple Kyu?
you have no idea what you are sayin.
Author:  Magicwand [ Thu Jul 21, 2016 9:07 pm ]
Post subject:  Re: Practical alternatives to superko
For those who think superko will work. I will give you simple triple ko. It has 6 move sequence to comeback to original positon.
imagine you have two triple ko. you can create 36 move sequence to make them come back to original position for two triple ko.
I am only talking about triple ko. now you imagine quad. i didnt do math on that but i will assume triple ko and quad ko.
will generat well over 100 move sequence to burnout possible move.

Now you KYU players who can not even understand simple life and death or review your own move past 10 move.
can you remember all 200 positions of ko if that come up? I strongly doubt that you remember past 10.

Now is that a solution for human being? NO!
Click Here To Show Diagram Code
[go]$$ Very simple triple ko w little twist to confuse you
$$ ---------------------------------------
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . , . . . . . , . . . . . , . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . , . . . . . , . . . . . , . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . X X X . |
$$ | . . . . . . . . . . . . . . X X . X X |
$$ | . . . . . . . . . . . . X X X O X O O |
$$ | . . . , . . . . . , . . X O O , O O X |
$$ | . . . . . . . . . . . . X X O O O X . |
$$ | . . . . . . . . . . . . . X O O X X X |
$$ | . . . . . . . . . . . . . X O X . X . |
$$ ---------------------------------------[/go]


In Korea when we play handycap game and score tie then it is a tie.
In Korean we call that (和局) <== I think it is correct character.
I think it means good game since nobody lost.
game doesnt necessary have winner and loser.
You will live to play another game.
Author:  luigi [ Thu Jul 21, 2016 10:47 pm ]
Post subject:  Re: Practical alternatives to superko
Magicwand wrote:
For those who think superko will work. I will give you simple triple ko. It has 6 move sequence to comeback to original positon.
imagine you have two triple ko. you can create 36 move sequence to make them come back to original position for two triple ko.
I am only talking about triple ko. now you imagine quad. i didnt do math on that but i will assume triple ko and quad ko.
will generat well over 100 move sequence to burnout possible move.

Now you KYU players who can not even understand simple life and death or review your own move past 10 move.
can you remember all 200 positions of ko if that come up? I strongly doubt that you remember past 10.

Now is that a solution for human being? NO!
Click Here To Show Diagram Code
[go]$$ Very simple triple ko w little twist to confuse you
$$ ---------------------------------------
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . , . . . . . , . . . . . , . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . , . . . . . , . . . . . , . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . X X X . |
$$ | . . . . . . . . . . . . . . X X . X X |
$$ | . . . . . . . . . . . . X X X O X O O |
$$ | . . . , . . . . . , . . X O O , O O X |
$$ | . . . . . . . . . . . . X X O O O X . |
$$ | . . . . . . . . . . . . . X O O X X X |
$$ | . . . . . . . . . . . . . X O X . X . |
$$ ---------------------------------------[/go]


This difficulty doesn't apply to the Basic-Fixed-Ko rules, though. You can see without much calculation that none of the groups can be killed. Therefore, since we know playing out cycles is never beneficial under those rules, none of the players will play in that area for the rest of the game (except perhaps for ko threats).
Author:  RobertJasiek [ Thu Jul 21, 2016 10:51 pm ]
Post subject:  Re: Practical alternatives to superko
For two triple or quadruple kos on the board, unless you know and can apply the theorem of Spight-Davies-Rickard (IIRC, these were involved) about how to fight multiple n-tuple-kos, the possible long sequences can indeed be (too) hard to remember, especially if the first few plays are played too quickly before really starting to try remembering moves.

Why care? In practical games, multiple n-tuple kos never occur! They are fun for theoreticians only. From a practical POV, the worst that would happen is one quadruple ko (or one quintuple ko) possibly with one most valuable basic ko fight elsewhere on the board. This is worth discussing because it occurs every ca. 50.000th game or less frequently.

One triple ko under superko, uh sure it is a 6 play cycle but emphasising this shows no understanding whatsoever how a triple ko can be fought under superko: like a basic ko. After one ko capture in the triple ko, make a ko threat elsewhere on the board. This is equivalent to three successive ko captures in the triple ko followed by a ko threat, which is equivalent to five successive ko captures in the triple ko followed by a ko threat. One triple ko is that easy!

Since the practically occurring cases can be handled by humans, superko is for humans.

Korean long cycle tie (like Japanese long cycle no result) has the same burden on the players to recognise occurrence of a cycle as superko, so is not easier at the moment of first creating a cycle. Only allowed recycling (for some time) without penalty can relax this so that stupid players can play cycles a few times before recognising that they are recycling indeed. Would they be proud of showing their weak cognitive skills?

Wait, if a game is a tie, then why play another game? Tournament rules exceptions, I see.

(Due to attending the European Go Congress, I might not find time to continue discussion until afterwards.)
Author:  RobertJasiek [ Thu Jul 21, 2016 10:55 pm ]
Post subject:  Re: Practical alternatives to superko
luigi wrote:
none of the groups can be killed


Black to move removes all the white stones;)
Author:  luigi [ Fri Jul 22, 2016 11:17 am ]
Post subject:  Re: Practical alternatives to superko
luigi wrote:
And, most importantly, groups involved in would-be cycles are just alive [under Basic-Fixed-Ko rules].

Whoops. On further thought, this is of course not quite true. In triple ko, for instance, one player will capture the opponent's group. The correct thing to say is that playing out a cycle will (most probably) never be better than not doing so.
Author:  Matti [ Tue Sep 13, 2016 5:10 am ]
Post subject:  Re: Practical alternatives to superko
oren wrote:

No result has been applied in real life many times. My favorite was a tv championship that ended that way. The players got a short break, change of officials and tv commentator, and they restarted the game.

That was easy...

For longer games, you can decide if you want ties or if you want a rescheduled match. Saying it's impractical is a bit silly.


I had once a dispute where my opponent took back his move and placed it on another intersection. There were no witnesses. The referee decided that must start a new game with the remaining time on our clocks. I had 6 minutes left and my opponent 9 minutes and we had 20 sec byoyomi. I won by half a point.
Author:  oren [ Tue Sep 13, 2016 8:58 am ]
Post subject:  Re: Practical alternatives to superko
Matti wrote:
I had once a dispute where my opponent took back his move and placed it on another intersection. There were no witnesses. The referee decided that must start a new game with the remaining time on our clocks. I had 6 minutes left and my opponent 9 minutes and we had 20 sec byoyomi. I won by half a point.


How would you have a takeback? For triple ko, you can call over the ref and continue from the current spot unless you disagree about where the very last move was.
Author:  Matti [ Thu Sep 15, 2016 4:39 am ]
Post subject:  Re: Practical alternatives to superko
oren wrote:
Matti wrote:
I had once a dispute where my opponent took back his move and placed it on another intersection. There were no witnesses. The referee decided that must start a new game with the remaining time on our clocks. I had 6 minutes left and my opponent 9 minutes and we had 20 sec byoyomi. I won by half a point.


How would you have a takeback? For triple ko, you can call over the ref and continue from the current spot unless you disagree about where the very last move was.


The takeback was not in ko fight. My point was that the new game, which could also occur after no result, was played with the remaining time in the clock, thus it did not create any problem with the schedule.
Author:  oren [ Thu Sep 15, 2016 1:13 pm ]
Post subject:  Re: Practical alternatives to superko
Matti wrote:
The takeback was not in ko fight. My point was that the new game, which could also occur after no result, was played with the remaining time in the clock, thus it did not create any problem with the schedule.


Sure, my preference is to handle it with a tie. Replays generally happen in professional tournaments with big money. In amateur events, I'd call it a tie and move on.
Author:  Matti [ Sun Sep 18, 2016 1:31 pm ]
Post subject:  Re: Practical alternatives to superko
oren wrote:
Matti wrote:
The takeback was not in ko fight. My point was that the new game, which could also occur after no result, was played with the remaining time in the clock, thus it did not create any problem with the schedule.


Sure, my preference is to handle it with a tie. Replays generally happen in professional tournaments with big money. In amateur events, I'd call it a tie and move on.

It is fine to have a tie, if it does not destroy the tournament system.
Page 2 of 2 All times are UTC - 8 hours [ DST ]
Powered by phpBB © 2000, 2002, 2005, 2007 phpBB Group
http://www.phpbb.com/